$0B lost to improper payments and fraud since 2003.

SPECTR catches the fraud and improper payments that legacy systems miss, built for the custodians of public money. Vendor disbursements: screen payments before auditors do. Grant programs: map California codes and federal rules into compliant applicant flows and audit-ready admin files.

01Two platforms. One payment integrity stack.

Payment Integrity

Catch the fraud and improper payments legacy systems miss.

Frontier-model ensembles and LLM reasoning across nine detection engines: deceased-recipient matching, behavioral drift, vendor clustering, contract intelligence. Built for teams screening grant disbursements and vendor payments — not application intake, not $50M sole-source bank contracts.

$236B
lost in FY2023
Explore

Policy Navigator

Turn buried grant and code requirements into compliant, audit-ready workflows.

California codes, HUD grant manuals, and federal regulations run hundreds of pages. Navigator extracts the rules into guided applications for constituents and structured review for admins — every answer tied to the source requirement, ready for monitoring and single audit.

PDF → Product
from code to compliant applications
Explore
02By the numbers

The problem isn't hiding. Almost nobody is looking for it. Grant and public fund programs are no exception.

The largest study of occupational fraud ever conducted, 2,402 real cases across 143 countries, shows the same pattern in every edition: organizations that monitor their own payments lose far less and catch fraud far faster. Almost none of them do.

3%

of occupational fraud is caught by automated data monitoring today. Tips catch 43%, usually after the money is gone.

53%

lower losses at organizations that monitor their payments proactively.

44%

faster detection when monitoring is proactive instead of passive.

12mo

median time a fraud scheme runs undetected before anyone notices.

$9,400

in added losses for every month a scheme stays undetected.

5%

of annual revenue lost to fraud by the typical organization.

Source: ACFE, Occupational Fraud 2026: A Report to the Nations (2,402 cases, 143 countries)
03Fraud in the news

These made headlines years too late. The evidence was there the whole time.

Every one of these schemes left a trail in the payment records before anyone noticed. Here's what happened, and the detection angle SPECTR would have used to flag it first.

Los Angeles Times
July 2010

Bell, CA: a city of 38,000 paid its administrator nearly $800K a year.

City officials in working-class Bell drew some of the highest municipal salaries in the nation, the city manager at roughly $800K and council members near $100K for part-time roles, buried in routine budget line items for years until reporters pulled the records.

How we'd catch it

Baseline drift on compensation vs. peer cities. Bell’s payroll-per-capita and executive pay sat multiples above comparable towns. A peer-normalized outlier check flags that gap on day one, not after a tip.

Outcome: Bell’s CAO sentenced to federal prison
Federal Bureau of Investigation
November 2012

Dixon, IL comptroller stole $53.7M, the largest municipal fraud in U.S. history.

Over two decades, Rita Crundwell, the comptroller of Dixon, Illinois (population about 16,000), siphoned $53.7 million of city money into a secret bank account she controlled, fabricating invoices and a fake state fund to cover the outflow while the small city cut budgets.

How we'd catch it

Vendor and cash-flow anomaly detection. Payments routing to an account that maps to no real vendor, and a single official’s transfers dwarfing the city’s actual budget, are textbook outliers our engines flag long before a 20-year run.

U.S. Attorney’s Office, District of Arizona
June 2025

Arizona county treasurer stole $38M over a decade and got 10 years.

Elizabeth Gutfahr, the elected Santa Cruz County treasurer from 2012 to 2024, wired roughly $38.7 million of county funds to shell companies she created across about 187 transfers, defeating the two-step approval control by using a subordinate’s security token to approve her own payments.

How we'd catch it

Dual-control and new-vendor anomaly detection. When the same person both initiates and approves wires, and the money lands at freshly created payees with no real business history, the self-approval loop and shell-account destinations stand out immediately.

U.S. Attorney’s Office, Central District of California
October 2024

School district finance chief embezzled $16.6M with checks to fake payees.

Jorge Contreras, senior director of fiscal services at Magnolia School District near Anaheim, cut more than 250 unauthorized district checks listing fictitious payees and deposited them into his own account, stealing about $16.6 million the district had hired him to protect.

How we'd catch it

Payee verification and ghost-vendor detection. Hundreds of checks to payees that match no verified vendor, all routing to one personal account, form a cluster our duplicate and behavioral engines surface before the next batch clears.

U.S. Attorney’s Office, Eastern District of California
September 2024

California schools official funneled $1.5M to a shell company, then bought a Ferrari.

Jeffrey Menge, former assistant superintendent of Patterson Joint Unified School District, and a former IT director routed about $1.5 million in district payments to CenCal Tech, a Nevada shell company Menge secretly controlled behind a fictitious executive named "Frank Barnes," spending proceeds on luxury cars.

How we'd catch it

Vendor legitimacy and beneficial-ownership cross-checks. A vendor with no real footprint, a fabricated officer, and a quiet tie to an insider surfaces under entity-resolution checks that match payees against ground-truth business records.

U.S. Department of Justice
December 2025

Two men drained a state labor agency of $17M in pandemic jobless benefits.

Malcolm Jeffrey and Gerard Towns of Cordele, Georgia were sentenced for defrauding the Georgia Department of Labor of at least $17 million, filing more than 2,500 fraudulent unemployment claims from 2020 to 2022 using fictitious employers and stolen identities.

How we'd catch it

Identity-graph and synthetic-employer detection. Bursts of claims sharing addresses, bank accounts, or device fingerprints, tied to employers with no real payroll history, cluster instantly under behavioral and identity analysis.

U.S. Attorney’s Office, Southern District of New York
February 2024

The largest single-day bribery takedown in DOJ history hit a public housing agency.

Seventy current and former New York City Housing Authority employees were charged with demanding over $2 million in bribes from contractors in exchange for steering more than $13 million in no-bid public housing repair work, typically taking 10 to 20 percent of each contract.

How we'd catch it

Procurement pattern analysis. No-bid awards concentrated among the same officials and contractors, sized just under approval thresholds, form a collusion signature our contract-intelligence engine maps across thousands of work orders.

U.S. Attorney’s Office, Southern District of Florida
May 2025

Miami police officer pleaded guilty to about $200K in pandemic loan fraud.

Tramaine Liptrot, a City of Miami police officer, pleaded guilty to wire fraud for fraudulently obtaining roughly $200,000 across two Paycheck Protection Program loans in 2020 and 2021 by submitting falsified payroll figures for a side business, including a bogus second-draw loan.

How we'd catch it

Cross-program and payroll-consistency checks. Relief applications with payroll figures that match no real filing, plus duplicate draws for the same entity, are flagged by cross-referencing every claim against ground-truth records.

U.S. Department of Justice
June 2025

DOJ charges half a billion dollars in healthcare and COVID fraud schemes.

A nationwide takedown charged defendants across schemes built on billing for services never rendered, ghost patients, and pandemic-relief abuse, the same patterns that quietly drain Medicare, Medicaid, and relief programs.

How we'd catch it

Duplicate-claim and deceased-recipient matching plus provider behavioral drift. Billing for the dead, the same service twice, or volumes no real practice could deliver are exactly the signals our ensembles surface before payment goes out.

04Built for a different problem

Most payment integrity tools are built for banks. Not for the agencies that need it most.

Existing fraud tools are built for banks with massive compliance teams and complex banking cores.



We protect public money, not bank profits.

Who it’s built for
Banks, payment processors, and fintechs with dedicated compliance teams.
Local agencies and governments — disbursement teams on vendor payments, compliance teams on codes and grant applications.
What it scores
One transaction at a time, as it happens.
Each vendor across its full history and nine engines, against the distribution of every legitimate peer.
How it learns
Supervised models that need a labeled history of past fraud to train on.
Geometric distribution analysis that works from day one.
How you deploy
A direct API line into a banking core. Months of integration.
Payment integrity: start with a CSV. Grant compliance: start with your code PDF.

We're not here to replace a fraud system you already run. For most of the agencies and governments we work with, there isn't one yet on disbursements — and grant staff are still reconciling applications against hundred-page code manuals by hand.

Free pilot

Send us your vendor payment CSV. We run the analysis for free. You get the report.

No procurement cycle to start. No integration project. Email us a vendor payment file, columns map automatically, and we send back a plain-English findings report with every flag explained and the math shown. If it's useful, we talk about a pilot. Grant compliance and audit-ready applications are Navigator — request a demo for that.

Email us your CSV, founders@spectr.llc

CSV stays yours · analysis runs on your file only · no data retained after the report